#编码

看到这个街头霸王，80 后的童年记忆回来了！ 确实很感慨，09 年毕业后刚学编程时，自己能写个扫雷、写的 1024 就好开心，写出来自己能玩半天 ... 那时记得还是 Flex 和 Silverlight，也是那时开始对前端特别感兴趣，一直到现在。 现在再回看，这些程序，基本都是 Claude Code / Codex 几分钟就能搞定的小玩意，甚至不咋需要思考！ 突然有个想法，我想把自己小时候和上学时觉得好玩的，现在已经失传的小游戏们，重新做一遍，也让儿子这个 10 后知道知道，他爸小时候有多开心！！ 先从我第一个手机 Moto E770 装的第一个 WAP 页游开始吧，记得叫做：打砖块？

I asked @bcherny how many Claude agents he's running right now as he sat next to me. "A few thousand." Boris (creator of Claude Code) came back to my Stanford course 𝗧𝗵𝗲 𝗠𝗼𝗱𝗲𝗿𝗻 𝗦𝗼𝗳𝘁𝘄𝗮𝗿𝗲 𝗗𝗲𝘃𝗲𝗹𝗼𝗽𝗲𝗿 last Thursday. His main takeaways: - 𝗔𝘂𝘁𝗼𝗺𝗮𝘁𝗲 𝘆𝗼𝘂𝗿 𝗯𝗼𝘁𝘁𝗹𝗲𝗻𝗲𝗰𝗸: Claude has written 100% of his team's code since November and output is up ~10x per engineer. So much code shipped that people stopped reading it carefully and started skimming, so they automated code review. Approvals are next. The team identified the main things impeding their development velocity and built systems to unblock themselves. - 𝗕𝘂𝗶𝗹𝗱 𝗳𝗼𝗿 𝘁𝗵𝗲 𝗺𝗼𝗱𝗲𝗹 𝟲 𝗺𝗼𝗻𝘁𝗵𝘀 𝗼𝘂𝘁: early Claude Code revenue was flat for months because the model wasn't smart enough yet. Then Opus 4 came out and growth became astronomic. His method is to read your agent transcripts and see where the model fails. That's where it will improve in the next model release. - 𝗦𝗺𝗮𝗹𝗹 𝘁𝗲𝗮𝗺𝘀 𝗴𝗼 𝗳𝗮𝘀𝘁𝗲𝗿: for months Claude Code was basically just him. With AI, one or two people can outship a team 100 times their size. Constraints breed creativity. - 𝗧𝗵𝗶𝗻𝗸 𝗶𝗻 𝗹𝗼𝗼𝗽𝘀: his grandpa programmed punch cards in the Soviet Union. Boris writes loops that run for days or weeks and spawn many agents. He expects that typing code by hand is heading where punch cards went. - 𝗗𝗼𝗺𝗮𝗶𝗻-𝘀𝗽𝗲𝗰𝗶𝗳𝗶𝗰 𝗯𝘂𝗶𝗹𝗱𝗲𝗿𝘀 𝗮𝗿𝗲 𝘁𝗵𝗲 𝗳𝘂𝘁𝘂𝗿𝗲: he expects the best accounting software of 2027 to be built by an accountant. When code is cheap, knowing the problem deeply is the only alpha In the 50 years after the printing press, Europe produced more literature than in the previous 1,000. He thinks coding is about to undergo a similar explosion of creation.

我靠，我靠, 我靠！兄弟们，我找到一个比菲律宾还便宜的订阅 20X 的地方。 Codex Pro 的 $200 月费档，GamsGo @GamsGo_Global 现在 $145.99/月。 这价格，比我在菲律宾区订阅的价位还要便宜了。 Codex、Claude、Grok 我自己都在订，这类工具用得多，看到 Pro 这档的价格，会特别留意一下。 刚看了 GamsGo 的 Codex 自营套餐：Pro-200，独享账号，1个月。 想再加一个 Codex 账号，或者准备上 Pro 的，这个价确实值得看看。 用评论区的链接下单，还可以再返 3%，相当于用U卡返现了。

2012 年：花 1 个月，用canvas写了个简单的街头霸王 2026 年：花 3 天，用 AI 做了个 3D 街头霸王，复杂代码一行都不用自己写了 🔗 试玩地址：https://street-fighter.tutti.sh/ 由于非常不擅长和没耐心做效果，主要用了这几个工具： 1. 混元 3D 建模，最开始试了下付费的 3D 建模工具（充了几十块只用了 1 次），发现混元效果更好一些，1 天 20 次免费，虽然只能去网页上用，不过让 agent 读一下网站自己的 cookie，然后直接让 agent 调就行。 2. Blender，做一些精修（实际也不知道修的怎么样，没太关注，就是让混元建模完之后修一下） codex和claude code，claude code 只有 20 美金的订阅，大部分活还是 codex 干的，偶尔也让 workbuddy 干一点 3. Tutti（我们自己开发的），主要用来协作和分享，负责建模和负责部署、音效的是另外的人，我主要写游戏逻辑，大家在同一个云端电脑环境里干活，实时协作性比较强。 全套提示词、过程都在：https://tutti.sh/share/v2/289b0c16-630d-43ae-91e0-b4a0b5932cd1?inviteType=guest&token=bjwqJx2VzR7scvtfwcBf1P_sJ3MZmQ0x7CoAPsS9HMM

智能体会写代码了，工程师还剩什么？ 吴恩达老师基于 1w+ 职位分析与数十次专家访谈，用五封公开信绘出「AI 工程技能地图」，答案分四层： 构建与部署不可预测的 AI 应用、守住软件工程根基、高效驾驭编程智能体，最终上移到塑造构建：决定该做什么，并对结果负责。 https://x.com/i/article/2098560800181125120

When our AI code reviewer's cost per run doubled, we dug into the token usage to find out why. Learn how we used prompt caching and intelligent model routing to slash our inference bill by 68% while speeding up reviews. https://cfl.re/4isj2NU

Introducing Flint, @Recursive_SI’s first model: a fast local coding model built with our automated research system, the Eureka Machine. • Best-in-class fast local coding • Up to 2x faster than comparable models: 125+ tokens/sec on a MacBook Pro, 150+ on RTX Spark • Built from scratch in under 4 months Private, unmetered intelligence on your own machine. Open weights available now on @huggingface, run it locally on your favorite framework or try it for free on @opencode or @pidotdev.

The Graphify skill by @graphify is now in Computer's skills marketplace, available to all Computer users. Turn a codebase into an interactive map of its files, functions, and dependencies. Tasks that use /graphify are free through October 31, up to 5,000 credits per user.

Introducing Flint, @Recursive_SI’s first model: a fast local coding model built with our automated research system, the Eureka Machine. • Best-in-class fast local coding • Up to 2x faster than comparable models: 125+ tokens/sec on a MacBook Pro, 150+ on RTX Spark • Built from scratch in under 4 months Private, unmetered intelligence on your own machine. Open weights available now on @huggingface, run it locally on your favorite framework or try it for free on @opencode or @pidotdev.

Introducing Flint, @Recursive_SI’s first model: a fast local coding model built with our automated research system, the Eureka Machine. • Best-in-class fast local coding • Up to 2x faster than comparable models: 125+ tokens/sec on a MacBook Pro, 150+ on RTX Spark • Built from scratch in under 4 months Private, unmetered intelligence on your own machine. Open weights available now on @huggingface, run it locally on your favorite framework or try it for free on @opencode or @pidotdev.

An update here: we’re upgrading all Codex for OSS grants to Pro 200! There will still be the 10,000 grants in this round, now at the higher tier. (If you've already received a grant, the upgrade should be reflected automatically.) Applications are still open :)

Introducing Flint, @Recursive_SI’s first model: a fast local coding model built with our automated research system, the Eureka Machine. • Best-in-class fast local coding • Up to 2x faster than comparable models: 125+ tokens/sec on a MacBook Pro, 150+ on RTX Spark • Built from scratch in under 4 months Private, unmetered intelligence on your own machine. Open weights available now on @huggingface, run it locally on your favorite framework or try it for free on @opencode or @pidotdev.

What is a piece of normal software that has improved the most because of its its developers using AI extensively? (Claude code doesn't count)

Inside Grok Bot: is the agent harness dead? @jediahkatz of @spacexai explains why smarter models let Cursor remove prompts and guardrails, and why Grok Bot is adding them back. Watch his AI Engineer New York talk: https://www.youtube.com/watch?v=7Y4sbZgBdnc

pstack v0.15.27 ships with a new skill /create-formal-verification-skill! it's still in beta, and similar to /create-verification-skill, this is a meta-skill that creates a new skill and formal verification suite based on the characteristics you want to verify. for example, you may want to model state machines and systems with concurrency using TLA+, or use kani in rust to check for undefined behavior. if you're new to formal verification, think of it as mathematically proving properties and invariants (something that must always be true) or "laws" of your system. usually, this means modeling your programs in a separate language like with TLA+ and then running the solver on that spec. newer approaches like with kani and @bendlang take a different approach where there is no separate language: you write kani in rust, and bend in bend (it's also a programming language, and the proofs are written as laws and expressed in its type system). historically formal verification never reached mainstream status because of the difficulty in modeling and writing proofs. it's quite difficult to model your business requirements and keep them in sync! but i expect that we will see more innovation in formal verification in the future as being able to formally verifiy that the code does what you intended gives your agents superpowers in writing correct software. in the meantime though, i've been experimenting with modeling our code with TLA+ and have already found several bugs with the skill created by /create-formal-verification-skill. try it out and let me know how it goes!

part 4 of "software is taking off" through recent discussions i realized many people saying "software is dead" were actually thinking about software engineering as a profession so.. is software engineering dead? when i was a kid, my mom was an accountant. when i was doing my homework in the evenings, I would hear the sound of her using abacus. it was very loud and noisy over the years, the sound of pushing abacus became punching calculators, then typing on keyboards, and eventually making phone calls. she just retired a few years ago, as a director of an accounting team if you thought of the accounting profession as "pushing abacus", it died a long time ago. but if you understood accountants' jobs are actually "helping businesses make the right financial decisions", you'll realize it'll never stop being relevant so whether software engineering is dead pretty much all depends on your mindset on what the profession even is if your definition is "i write code others tell me to write", yup that's already dead if it's some form of "i write very good code", it's not quite dead yet, but i speculate it'll be dead within the next 1-3 years. i'll link in the thread below another post where i explained why so what's the type of engineer that's seeing growing demand? fig. 1 shows SWE overall seeing a growth in job posting in the US. it tells quite a story when that's happening during a year when coding agents have seen the craziest advancements and adoption fig. 2 shows a lot of the growth came from senior, or AI related roles. i will touch on why below fig. 3 shows the rise of FDEs the growing demand for software engineers is for those who has the right capability to solve problems with high precision, agency, and leverage - these traits have always been desired long before AI came along, and are seeing no sign of stopping just think - when two competing companies both have access to the same AI, who wins? it's going to be their humans that make the difference, and they are going to get as many of such humans as they can the software engineering profession has never been about writing punch cards, writing assembly, or writing C. it's never been about using vim, emacs, or visual studio. the code, the tools, and the activities have always been changing nonstop the profession has always been about "solving problems with software". now there are a lot more problems that just became solvable, and everyone has gained quite an insane amount of leverage - there's never been a better time

Over 1 million lines of dead code removed. 🧹 Meet AutoClean, Uber's AI-powered system that's cleaned up 16,000+ stale feature flags using AI agents, deterministic code transformations, and validation. https://x.com/i/article/2112273498110660608